Импорт библиотек¶
In [1]:
import warnings
warnings.filterwarnings('ignore')
import os
import io
import json
import shutil
import zipfile
import requests
import numpy as np
import pandas as pd
import seaborn as sns
import tensorflow as tf
from tqdm.auto import trange
import matplotlib.pyplot as plt
from scipy.signal import welch
from scipy.stats import kurtosis, skew
from scipy.signal.windows import blackman
from sklearn.model_selection import train_test_split, GridSearchCV
from sklearn.preprocessing import PowerTransformer, StandardScaler, MinMaxScaler
from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score, classification_report, confusion_matrix, ConfusionMatrixDisplay, roc_curve, auc
from sklearn.svm import SVC
from sklearn.naive_bayes import GaussianNB
from sklearn.tree import DecisionTreeClassifier
from sklearn.neural_network import MLPClassifier
from sklearn.neighbors import KNeighborsClassifier
from sklearn.linear_model import LogisticRegression
from xgboost import XGBClassifier
from lightgbm import LGBMClassifier
from catboost import CatBoostClassifier
from sklearn.ensemble import RandomForestClassifier, AdaBoostClassifier, GradientBoostingClassifier, HistGradientBoostingClassifier, StackingClassifier, VotingClassifier
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Dropout
from tensorflow.keras.callbacks import EarlyStopping
from tensorflow.keras.utils import to_categorical, plot_model
Загрузка и первичная обработка данных¶
Загрузка данных с GitHub¶
In [2]:
url = "https://github.com/tiiuae/UAV-Realistic-Fault-Dataset/archive/refs/heads/main.zip"
response = requests.get(url)
zip_file = zipfile.ZipFile(io.BytesIO(response.content))
zip_file.extractall("UAV-Realistic-Fault-Dataset")
Парсинг данных¶
In [3]:
base_dir = r'UAV-Realistic-Fault-Dataset\UAV-Realistic-Fault-Dataset-main\Dataset'
processed_dir = 'ProcessedDataset'
os.makedirs(processed_dir, exist_ok=True)
for folder in trange(5, desc='Progress', colour='blue'):
folder_path = os.path.join(base_dir, str(folder))
file_number = 0
class_folder = os.path.join(processed_dir, str(folder))
os.makedirs(class_folder, exist_ok=True)
for subfolder in os.listdir(folder_path):
subfolder_path = os.path.join(folder_path, subfolder)
if os.path.isdir(subfolder_path):
combined_data = []
for file in os.listdir(subfolder_path):
if file.endswith('SensorCombined.jsonl'):
with open(os.path.join(subfolder_path, file), 'r') as jsonl_file:
for line in jsonl_file:
data = json.loads(line)
combined_data.append(data)
df = pd.DataFrame(combined_data)
for i, row in enumerate(df['gyro_rad']):
df.loc[i, ['gx', 'gy', 'gz']] = row
for i, row in enumerate(df['accelerometer_m_s2']):
df.loc[i, ['ax', 'ay', 'az']] = row
df.drop(columns=["gyro_rad", "accelerometer_m_s2",
"accelerometer_timestamp_relative",
"accelerometer_clipping"], inplace=True)
df.rename(columns={'timestamp': 'time',
'gyro_integral_dt': 'gInt',
'accelerometer_integral_dt': 'aInt'}, inplace=True)
df.to_csv(os.path.join(class_folder, f"class_{folder}_number_{file_number}.csv"), index=False)
file_number += 1
combined_class_data = []
for file in os.listdir(class_folder):
if file.endswith('.csv'):
file_path = os.path.join(class_folder, file)
class_df = pd.read_csv(file_path)
combined_class_data.append(class_df)
if combined_class_data:
final_df = pd.concat(combined_class_data, ignore_index=True)
final_df.to_csv(os.path.join(class_folder, f"data_{folder}.csv"), index=False)
Progress: 0%| | 0/5 [00:00<?, ?it/s]
In [4]:
# Убедимся, что данные корректно записались
test_data = pd.read_csv(r"ProcessedDataset\0\data_0.csv")
test_data.head(5)
Out[4]:
| time | gInt | aInt | gx | gy | gz | ax | ay | az | |
|---|---|---|---|---|---|---|---|---|---|
| 0 | 947975062 | 4998 | 4998 | 0.008529 | 0.004348 | -0.001269 | 0.145165 | -0.034807 | -9.828311 |
| 1 | 947976312 | 4996 | 4996 | 0.001431 | 0.005915 | -0.001310 | 0.116024 | 0.020439 | -9.769987 |
| 2 | 947983469 | 4998 | 4998 | 0.016173 | 0.007350 | 0.000573 | 0.092866 | 0.060234 | -9.827328 |
| 3 | 947996546 | 4994 | 4994 | -0.001588 | 0.000100 | -0.003505 | 0.065213 | 0.016342 | -9.804899 |
| 4 | 947995456 | 4995 | 4995 | 0.008024 | 0.012801 | 0.000042 | 0.082880 | 0.064342 | -9.800008 |
Удаление ненужных данных¶
In [5]:
folder_path = 'UAV-Realistic-Fault-Dataset'
if os.path.exists(folder_path):
shutil.rmtree(folder_path)
Анализ данных¶
Вычисление частоты дискретизации¶
Частота дискретизации акселерометра и гироскопа нигде не указана, поэтому вычислим её самостоятельно.
In [6]:
lst = []
for clas in range(5):
for num in range(20):
if clas == 0 and num == 19: continue # Так как в классе "0" меньше на 1 файл
data = pd.read_csv(rf"ProcessedDataset\{clas}\class_{clas}_number_{num}.csv") # Считываем данные
data = data.sort_values(by='time') # Сортируем по времени
time_differences = (data['time'].diff().abs()).dropna() # Рассчитываем разницу между всеми соседними
average_difference = time_differences.mean() / 10**6 # Переводим в секунды из микросекунд
lst.append(1 / average_difference) # Сохраняем
print(f"Рассчитанная частота дискретизации: {round(sum(lst) / len(lst))}")
Рассчитанная частота дискретизации: 195
Построение 3D спектрограмм¶
In [7]:
def plot_3d_spectrogram(clas, window_size=256):
"""
Строит 3D спектрограмму для заданного класса данных.
Параметры:
clas (str): Имя класса, для которого будет построена спектрограмма.
window_size (int, optional): Размер окна для обработки данных.
По умолчанию 256.
Возвращает:
None: Функция отображает графики, но не возвращает никаких значений.
Описание:
Функция считывает данные из CSV файла, применяет оконную функцию Блэкмана-Наталла,
выполняет быстрое преобразование Фурье (БПФ) для перекрывающихся сегментов данных
и строит 3D спектрограммы для каждого канала данных (gx, gy, gz, ax, ay, az).
Каждая спектрограмма отображает амплитуды в зависимости от частоты и времени.
"""
channels = ["gx", "gy", "gz", "ax", "ay", "az"]
data = pd.read_csv(rf"ProcessedDataset\{clas}\data_{clas}.csv")[["time", *channels]].values
data[:, 6] += 9.81
Fs = 195 # Частота дискретизации
N = len(data) # Количество строк данных
overlap = window_size // 2 # Наложение окон (50%)
num_windows = (N - overlap) // (window_size - overlap) # Количество всех окон
amplitude = np.empty((num_windows, window_size, 6)) # Массив для амплитуд
xf = np.fft.fftfreq(window_size, 1 / Fs) # Получение частот для спектра
window = blackman(window_size) # Окно Блэкмана-Наталла
for i in range(num_windows):
# Разбивка данных на перекрывающиеся сегменты
start = i * (window_size - overlap)
end = start + window_size
segment = data[start:end, :]
# Сортировка по времени и проверка "стыков" между разными фрагментами данных
segment = pd.DataFrame(segment, columns=["time", *channels]).sort_values(by='time')
time_differences = (segment['time'].diff().abs()).dropna()
if max(time_differences) > 150000: continue
segment = segment.drop(columns=["time"]).values
if segment.shape[0] == window_size:
segment_windowed = segment * window[:, None] # Наложение окна
yf = np.fft.fft(segment_windowed, axis=0) # Применение БПФ
amplitude[i] = np.abs(yf) # Извлечение амплитуд
# Отрисовка графиков
fig = plt.figure(figsize=(20, 14))
for i in range(6):
ax = fig.add_subplot(2, 3, i + 1, projection='3d')
time_axis = np.arange(num_windows)
frequency_axis = xf[:window_size // 2]
amplitude_channel = amplitude[:, :window_size // 2, i]
time_axis_grid, frequency_axis_grid = np.meshgrid(time_axis, frequency_axis)
ax.plot_surface(time_axis_grid, frequency_axis_grid, amplitude_channel.T, cmap='viridis')
ax.set_title(f'3D Спектрограмма для {channels[i]}, class {clas}', size=12)
ax.set_xlabel('Номер окна')
ax.set_ylabel('Частота (Гц)')
ax.set_zlabel('Амплитуда')
# plt.savefig(f'3D_spectrogram_class_{clas}.png', dpi=300, bbox_inches='tight')
plt.show()
In [8]:
for i in range(5): plot_3d_spectrogram(i)
Извлечение признаков¶
In [9]:
# ЧО - частотная область, ВО - временная область
cols = ["ЧО Медиана", "ЧО Средняя",
"ЧО Std", "ЧО Max", "ЧО Min",
"ЧО 90 процентиль", "ЧО 75 процентиль", "ЧО 25 процентиль",
"ЧО Куртозис", "ЧО Ассиметрия", "ЧО Энергия", "ЧО Вариация",
"ЧО Количество пиков",
"ЧО Средняя > 10 Гц", "ЧО Std > 10 Гц", "ЧО Энергия > 10 Гц",
"ВО Медиана", "ВО Средняя",
"ВО Std", "ВО Max", "ВО Min",
"ВО 90 процентиль", "ВО 75 процентиль", "ВО 25 процентиль",
"ВО Куртозис", "ВО Ассиметрия", "ВО Энергия", "ВО Вариация",
"ВО Количество пиков",
"ВО Общая мощность", "ВО Средняя мощность"]
channels = ["gx", "gy", "gz", "ax", "ay", "az"]
cols = sum([[col[:3] + name + col[2:] for col in cols] for name in channels], [])
cols.append("Класс")
features_df = pd.DataFrame(columns=cols)
Fs = 195 # Частота дискретизации
window_size = 256 # Размер окна
overlap = window_size // 2 # Наложение окон (50%)
window = blackman(window_size) # Окно Блэкмана-Наталла
more10hz = round(256 / Fs * 10) # Для частот более 10 Гц
frequencies = np.fft.fftfreq(window_size, 1 / Fs) # Получение частот для спектра
for clas in trange(5, desc='Progress', colour='blue'):
data = pd.read_csv(rf"ProcessedDataset\{clas}\data_{clas}.csv")[["time", *channels]].values
num_windows = (len(data) - overlap) // (window_size - overlap) # Количество всех окон
for i in trange(num_windows, desc='Progress', colour='green'):
# Разбивка данных на перекрывающиеся сегменты
start = i * (window_size - overlap)
end = start + window_size
segment = data[start:end, :]
# Сортировка по времени и проверка "стыков" между разными фрагментами данных
segment = pd.DataFrame(segment, columns=["time", *channels]).sort_values(by='time')
time_differences = (segment['time'].diff().abs()).dropna()
if max(time_differences) > 150000: continue
segment = segment.drop(columns=["time"]).values
# Наложение окна, применение БПФ и извлечение признаков
if segment.shape[0] == window_size:
segment_windowed = segment * window[:, None] # Наложение окна
yf = np.fft.fft(segment_windowed, axis=0) # Применение БПФ
amplitude = np.abs(yf) # Извлечение амплитуд # shape = (num_windows, window_size, 6)
features = []
for channel in range(amplitude.shape[1]):
amp_half = amplitude[:overlap, channel] # Берем только положительные частоты
seg = segment[:, channel]
# Частотный спектр
features.append(np.median(amp_half)) # Медиана
features.append(np.mean(amp_half)) # Средняя амплитуда
features.append(np.std(amp_half)) # Стандартное отклонение
features.append(np.max(amp_half)) # Максимальная амплитуда
features.append(np.min(amp_half)) # Минимальная амплитуда
features.append(np.percentile(amp_half, 90)) # 90-й процентиль
features.append(np.percentile(amp_half, 75)) # 75-й процентиль
features.append(np.percentile(amp_half, 25)) # 25-й процентиль
features.append(kurtosis(amp_half)) # Куртозис
features.append(skew(amp_half)) # Ассиметрия
features.append(np.trapz(amp_half ** 2)) # Энергия
# Коэффициент вариации
features.append(np.std(amp_half) / np.mean(amp_half))
# Количество пиков (выбросов)
features.append(np.sum(amp_half > np.median(amp_half) + 0.5 * np.std(amp_half)))
# Исследование частот > 10 Гц
amp_half_10 = amp_half[more10hz:]
features.append(np.mean(amp_half_10)) # Средняя амплитуда частот > 10 Гц
features.append(np.std(amp_half_10)) # Стандартное отклонение частот > 10 Гц
features.append(np.trapz(amp_half_10 ** 2)) # Энергия частот > 10 Гц
# Временная область
features.append(np.median(seg)) # Медиана
features.append(np.mean(seg)) # Средняя амплитуда
features.append(np.std(seg)) # Стандартное отклонение
features.append(np.max(seg)) # Максимальная амплитуда
features.append(np.min(seg)) # Минимальная амплитуда
features.append(np.percentile(seg, 90)) # 90-й процентиль
features.append(np.percentile(seg, 75)) # 75-й процентиль
features.append(np.percentile(seg, 25)) # 25-й процентиль
features.append(kurtosis(seg)) # Куртозис
features.append(skew(seg)) # Ассиметрия
features.append(np.trapz(seg ** 2)) # Энергия
# Коэффициент вариации
features.append(np.std(seg) / np.mean(seg))
# Количество пиков (выбросов)
features.append(np.sum(seg > np.median(seg) + 0.5 * np.std(seg)))
# Плотность мощности
f, Pxx = welch(seg, fs=Fs, window='hann', nperseg=window_size, noverlap=overlap)
features.append(np.sum(Pxx[:overlap])) # Общая мощность
features.append(np.mean(Pxx[:overlap])) # Средняя мощность
# Добавляем метку класса и сохраняем извлеченные признаки
features.append(clas)
features_df.loc[len(features_df)] = features
features_df.to_csv(f"features.csv", index=False)
Progress: 0%| | 0/5 [00:00<?, ?it/s]
Progress: 0%| | 0/2623 [00:00<?, ?it/s]
Progress: 0%| | 0/2896 [00:00<?, ?it/s]
Progress: 0%| | 0/2817 [00:00<?, ?it/s]
Progress: 0%| | 0/2846 [00:00<?, ?it/s]
Progress: 0%| | 0/2803 [00:00<?, ?it/s]
Построение, тестирование и сравнение моделей¶
In [12]:
# Загрузка данных
features_df = pd.read_csv('features.csv')
X = features_df.drop(columns = ["Класс"])
y = features_df["Класс"].values
# Разделение на обучающую и тестовую выборки
random_state = 42
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=random_state)
# Приведение к нормальному распределению (Преобразование Yeo-Johnson)
power_transformer = PowerTransformer(method='yeo-johnson')
X_train = power_transformer.fit_transform(X_train)
X_test = power_transformer.transform(X_test)
# Стандартизация данных
standard_scaler = StandardScaler()
X_train = standard_scaler.fit_transform(X_train)
X_test = standard_scaler.transform(X_test)
# Нормализация данных
scaler = MinMaxScaler()
X_train = scaler.fit_transform(X_train)
X_test = scaler.transform(X_test)
# Инициализация моделей
models = {
"K-Nearest Neighbors": KNeighborsClassifier(n_neighbors=4, leaf_size=1, p=1, metric='manhattan', weights='distance'),
"Naive Bayes": GaussianNB(var_smoothing=0.1),
"Support Vector Machine": SVC(C=800, gamma=0.1, class_weight='balanced', tol=0.01, probability=True, random_state=random_state),
"Logistic Regression": LogisticRegression(C=300, solver='newton-cg', random_state=random_state),
"Decision Tree": DecisionTreeClassifier(max_depth=30, criterion='entropy', random_state=random_state),
"Perceptron": MLPClassifier(hidden_layer_sizes=(144, 256, 64, 16, 5), max_iter=400, random_state=random_state),
"Random Forest": RandomForestClassifier(n_estimators=400, max_depth=20, random_state=random_state),
"CatBoost": CatBoostClassifier(silent=True, random_state=random_state),
"AdaBoost": AdaBoostClassifier(estimator=DecisionTreeClassifier(max_depth=6, random_state=random_state), n_estimators=400, learning_rate=0.1),
"Gradient Boosting": GradientBoostingClassifier(random_state=random_state),
"Histogram-based Gradient Boosting": HistGradientBoostingClassifier(random_state=random_state),
"XGBoost": XGBClassifier(random_state=random_state),
"LightGBM": LGBMClassifier(verbose=-1, random_state=random_state),
"Stacking Classifier": StackingClassifier(
estimators=[("LGBMClassifier", LGBMClassifier(verbose=-1, random_state=random_state)),
("Random Forest", RandomForestClassifier(n_estimators=400, max_depth=20, random_state=random_state)),
("XGBClassifier", XGBClassifier(random_state=random_state)),
("CatBoost", CatBoostClassifier(silent=True, random_state=random_state)),
("Histogram-based Gradient Boosting", HistGradientBoostingClassifier(random_state=random_state)),
("K-Nearest Neighbors", KNeighborsClassifier(n_neighbors=4, leaf_size=1, p=1, metric='manhattan', weights='distance')),
("Support Vector Machine", SVC(C=800, gamma=0.1, class_weight='balanced', tol=0.01, probability=True, random_state=random_state))],
final_estimator=LogisticRegression(random_state=random_state)),
"Voting Classifier": VotingClassifier(
estimators=[("LGBMClassifier", LGBMClassifier(verbose=-1, random_state=random_state)),
("Random Forest", RandomForestClassifier(n_estimators=400, max_depth=20, random_state=random_state)),
("XGBClassifier", XGBClassifier(random_state=random_state)),
("CatBoost", CatBoostClassifier(silent=True, random_state=random_state)),
("Histogram-based Gradient Boosting", HistGradientBoostingClassifier(random_state=random_state)),
("K-Nearest Neighbors", KNeighborsClassifier(n_neighbors=4, leaf_size=1, p=1, metric='manhattan', weights='distance')),
("Support Vector Machine", SVC(C=800, gamma=0.1, class_weight='balanced', tol=0.01, probability=True, random_state=random_state)),
("Logistic Regression", LogisticRegression(C=300, solver='newton-cg', random_state=random_state)),
("Perceptron", MLPClassifier(hidden_layer_sizes=(144, 256, 64, 16, 5), max_iter=400, random_state=random_state))],
voting='soft')
}
# Обучение и тестирование моделей
for model_name, model in models.items():
model.fit(X_train, y_train)
y_pred = model.predict(X_test)
print(f"\n{model_name}:\n")
print(classification_report(y_test, y_pred))
print(f"Accuracy: {round(accuracy_score(y_test, y_pred) * 100, 2)} %\n", )
print("-" * 53)
K-Nearest Neighbors:
precision recall f1-score support
0.0 0.98 0.98 0.98 528
1.0 0.96 0.97 0.96 561
2.0 0.93 0.92 0.92 538
3.0 0.89 0.89 0.89 582
4.0 0.91 0.91 0.91 550
accuracy 0.93 2759
macro avg 0.93 0.93 0.93 2759
weighted avg 0.93 0.93 0.93 2759
Accuracy: 93.29 %
-----------------------------------------------------
Naive Bayes:
precision recall f1-score support
0.0 0.63 1.00 0.77 528
1.0 0.96 0.80 0.87 561
2.0 0.87 0.50 0.64 538
3.0 0.62 0.62 0.62 582
4.0 0.75 0.77 0.76 550
accuracy 0.74 2759
macro avg 0.77 0.74 0.73 2759
weighted avg 0.77 0.74 0.73 2759
Accuracy: 73.69 %
-----------------------------------------------------
Support Vector Machine:
precision recall f1-score support
0.0 0.93 0.95 0.94 528
1.0 0.95 0.95 0.95 561
2.0 0.91 0.91 0.91 538
3.0 0.89 0.88 0.89 582
4.0 0.91 0.90 0.91 550
accuracy 0.92 2759
macro avg 0.92 0.92 0.92 2759
weighted avg 0.92 0.92 0.92 2759
Accuracy: 91.88 %
-----------------------------------------------------
Logistic Regression:
precision recall f1-score support
0.0 0.92 0.91 0.92 528
1.0 0.90 0.92 0.91 561
2.0 0.81 0.84 0.83 538
3.0 0.84 0.82 0.83 582
4.0 0.85 0.84 0.85 550
accuracy 0.87 2759
macro avg 0.87 0.87 0.87 2759
weighted avg 0.87 0.87 0.87 2759
Accuracy: 86.59 %
-----------------------------------------------------
Decision Tree:
precision recall f1-score support
0.0 0.97 0.97 0.97 528
1.0 0.95 0.96 0.95 561
2.0 0.89 0.89 0.89 538
3.0 0.88 0.84 0.86 582
4.0 0.89 0.91 0.90 550
accuracy 0.91 2759
macro avg 0.91 0.91 0.91 2759
weighted avg 0.91 0.91 0.91 2759
Accuracy: 91.3 %
-----------------------------------------------------
Perceptron:
precision recall f1-score support
0.0 0.98 0.88 0.93 528
1.0 0.70 0.97 0.81 561
2.0 0.88 0.87 0.87 538
3.0 0.89 0.78 0.83 582
4.0 0.95 0.81 0.87 550
accuracy 0.86 2759
macro avg 0.88 0.86 0.86 2759
weighted avg 0.88 0.86 0.86 2759
Accuracy: 86.08 %
-----------------------------------------------------
Random Forest:
precision recall f1-score support
0.0 0.99 0.99 0.99 528
1.0 0.98 0.99 0.98 561
2.0 0.95 0.95 0.95 538
3.0 0.93 0.91 0.92 582
4.0 0.95 0.95 0.95 550
accuracy 0.96 2759
macro avg 0.96 0.96 0.96 2759
weighted avg 0.96 0.96 0.96 2759
Accuracy: 95.87 %
-----------------------------------------------------
CatBoost:
precision recall f1-score support
0.0 1.00 0.99 0.99 528
1.0 0.97 0.98 0.98 561
2.0 0.95 0.95 0.95 538
3.0 0.94 0.93 0.93 582
4.0 0.96 0.95 0.96 550
accuracy 0.96 2759
macro avg 0.96 0.96 0.96 2759
weighted avg 0.96 0.96 0.96 2759
Accuracy: 96.16 %
-----------------------------------------------------
AdaBoost:
precision recall f1-score support
0.0 0.99 0.98 0.99 528
1.0 0.97 0.98 0.98 561
2.0 0.87 0.95 0.91 538
3.0 0.78 0.89 0.83 582
4.0 0.98 0.74 0.84 550
accuracy 0.91 2759
macro avg 0.92 0.91 0.91 2759
weighted avg 0.92 0.91 0.91 2759
Accuracy: 90.72 %
-----------------------------------------------------
Gradient Boosting:
precision recall f1-score support
0.0 0.99 0.98 0.98 528
1.0 0.96 0.97 0.96 561
2.0 0.93 0.93 0.93 538
3.0 0.90 0.91 0.90 582
4.0 0.95 0.93 0.94 550
accuracy 0.94 2759
macro avg 0.94 0.94 0.94 2759
weighted avg 0.94 0.94 0.94 2759
Accuracy: 94.35 %
-----------------------------------------------------
Histogram-based Gradient Boosting:
precision recall f1-score support
0.0 0.99 0.99 0.99 528
1.0 0.98 0.98 0.98 561
2.0 0.95 0.96 0.95 538
3.0 0.94 0.93 0.93 582
4.0 0.95 0.95 0.95 550
accuracy 0.96 2759
macro avg 0.96 0.96 0.96 2759
weighted avg 0.96 0.96 0.96 2759
Accuracy: 96.23 %
-----------------------------------------------------
XGBoost:
precision recall f1-score support
0.0 1.00 0.99 0.99 528
1.0 0.98 0.99 0.98 561
2.0 0.96 0.95 0.96 538
3.0 0.93 0.93 0.93 582
4.0 0.95 0.95 0.95 550
accuracy 0.96 2759
macro avg 0.96 0.96 0.96 2759
weighted avg 0.96 0.96 0.96 2759
Accuracy: 96.27 %
-----------------------------------------------------
LightGBM:
precision recall f1-score support
0.0 0.99 0.99 0.99 528
1.0 0.98 0.98 0.98 561
2.0 0.95 0.96 0.95 538
3.0 0.93 0.93 0.93 582
4.0 0.95 0.95 0.95 550
accuracy 0.96 2759
macro avg 0.96 0.96 0.96 2759
weighted avg 0.96 0.96 0.96 2759
Accuracy: 96.19 %
-----------------------------------------------------
Stacking Classifier:
precision recall f1-score support
0.0 0.99 0.99 0.99 528
1.0 0.98 0.99 0.98 561
2.0 0.96 0.96 0.96 538
3.0 0.94 0.94 0.94 582
4.0 0.96 0.96 0.96 550
accuracy 0.97 2759
macro avg 0.97 0.97 0.97 2759
weighted avg 0.97 0.97 0.97 2759
Accuracy: 96.59 %
-----------------------------------------------------
Voting Classifier:
precision recall f1-score support
0.0 0.99 0.99 0.99 528
1.0 0.98 0.99 0.98 561
2.0 0.96 0.96 0.96 538
3.0 0.93 0.94 0.93 582
4.0 0.96 0.95 0.96 550
accuracy 0.96 2759
macro avg 0.97 0.97 0.97 2759
weighted avg 0.96 0.96 0.96 2759
Accuracy: 96.48 %
-----------------------------------------------------
In [13]:
# Сравнение моделей по метрике точности accuracy
models_accuracy_dict = {}
for name, model in models.items():
accuracy = accuracy_score(y_test, model.predict(X_test))
models_accuracy_dict[name] = round(accuracy * 100, 3)
models_accuracy_df = pd.DataFrame({
'Модель': models_accuracy_dict.keys(),
'Точность': models_accuracy_dict.values()
})
models_accuracy_df = models_accuracy_df.sort_values(by='Точность', ascending=False)
plt.figure(figsize=(8, 6))
plt.barh(models_accuracy_df['Модель'], models_accuracy_df['Точность'], color='skyblue')
plt.xlabel('Точность (%)')
plt.title('Сравнение точности моделей')
plt.gca().invert_yaxis()
plt.grid(axis='x')
# plt.savefig('models_accuracy_comparison.png', dpi=1200, bbox_inches='tight')
plt.show()
In [14]:
# Оценка моделей по четырем метрикам точности
models_acc_dict = {}
for name, model in models.items():
y_pred = model.predict(X_test)
accuracy = accuracy_score(y_test, y_pred)
precision = precision_score(y_test, y_pred, average='weighted')
recall = recall_score(y_test, y_pred, average='weighted')
f1 = f1_score(y_test, y_pred, average='weighted')
models_acc_dict[name] = {
'Accuracy': round(accuracy * 100, 3),
'Precision': round(precision * 100, 3),
'Recall': round(recall * 100, 3),
'F1 Score': round(f1 * 100, 3),
}
models_acc_df = pd.DataFrame(models_acc_dict).T
models_acc_df = models_acc_df.sort_values(by='Accuracy', ascending=False)
models_acc_df
Out[14]:
| Accuracy | Precision | Recall | F1 Score | |
|---|---|---|---|---|
| Stacking Classifier | 96.593 | 96.595 | 96.593 | 96.593 |
| Voting Classifier | 96.484 | 96.486 | 96.484 | 96.484 |
| XGBoost | 96.267 | 96.270 | 96.267 | 96.267 |
| Histogram-based Gradient Boosting | 96.231 | 96.229 | 96.231 | 96.230 |
| LightGBM | 96.194 | 96.199 | 96.194 | 96.197 |
| CatBoost | 96.158 | 96.155 | 96.158 | 96.154 |
| Random Forest | 95.868 | 95.857 | 95.868 | 95.860 |
| Gradient Boosting | 94.346 | 94.354 | 94.346 | 94.347 |
| K-Nearest Neighbors | 93.295 | 93.292 | 93.295 | 93.291 |
| Support Vector Machine | 91.881 | 91.874 | 91.881 | 91.874 |
| Decision Tree | 91.301 | 91.281 | 91.301 | 91.280 |
| AdaBoost | 90.721 | 91.589 | 90.721 | 90.696 |
| Logistic Regression | 86.589 | 86.625 | 86.589 | 86.592 |
| Perceptron | 86.082 | 87.877 | 86.082 | 86.311 |
| Naive Bayes | 73.686 | 76.717 | 73.686 | 73.304 |
In [15]:
# Построение матрицы ошибок для Stacking Classifier
plt.figure(figsize=(6, 5))
sns.heatmap(confusion_matrix(y_test, models.get("Stacking Classifier").predict(X_test)),
annot=True,
cmap=sns.color_palette("RdYlGn", n_colors=1000),
center=0,
fmt='d')
plt.title('Матрица ошибок для Stacking Classifier')
# plt.savefig('stacking_сlassifier_confusion_matrix.png', dpi=1200, bbox_inches='tight')
plt.show()
In [16]:
# Расчет точности с учетом ошибки в классе не более чем на 1
y_pred = models.get("Stacking Classifier").predict(X_test)
correct_predictions = 0
for true_class, predicted_class in zip(y_test, y_pred):
if abs(true_class - predicted_class) <= 1:
correct_predictions += 1
new_accuracy = correct_predictions / len(y_test)
print(f"Точность с учетом ошибки в классе не более чем на 1: {round(new_accuracy * 100, 2)} %")
Точность с учетом ошибки в классе не более чем на 1: 99.71 %
In [17]:
# Построение ROC-кривой для Stacking Classifier
voting_model = models["Stacking Classifier"]
y_score = voting_model.predict_proba(X_test)
classes = np.unique(y)
# Построение ROC-кривой для каждого класса
plt.figure(figsize=(10, 8))
for i in range(len(classes)):
fpr, tpr, _ = roc_curve(y_test, y_score[:, i], pos_label=classes[i])
roc_auc = auc(fpr, tpr)
plt.plot(fpr, tpr, label=f'ROC curve for class {classes[i]} (area = {roc_auc:.5f})')
plt.plot([0, 1], [0, 1], 'k--')
plt.xlim([0.0, 1.0])
plt.ylim([0.0, 1.05])
plt.xlabel('False Positive Rate')
plt.ylabel('True Positive Rate')
plt.title('Receiver Operating Characteristic (ROC) Curve')
plt.legend(loc='lower right')
plt.grid()
# plt.savefig('stacking_classifier_roc_curve.png', dpi=1200, bbox_inches='tight')
plt.show()
Пример использования GridSearchCV для поиска оптимальных параметров¶
In [18]:
param_grid = {
'n_neighbors': [1, 2, 3, 4, 5],
'leaf_size': [1, 2, 3],
'metric': ['manhattan', 'euclidean'],
'weights': ['uniform', 'distance'],
'p': [1, 2],
}
grid_search = GridSearchCV(estimator=KNeighborsClassifier(), param_grid=param_grid, cv=5, scoring='accuracy', n_jobs=-1)
grid_search.fit(X_train, y_train)
print("Лучшие параметры: ", grid_search.best_params_)
print("Лучшая оценка перекрестной проверки: ", grid_search.best_score_)
Лучшие параметры: {'leaf_size': 1, 'metric': 'manhattan', 'n_neighbors': 3, 'p': 1, 'weights': 'distance'}
Лучшая оценка перекрестной проверки: 0.9236128622367566
Создание, обучение и тестирование нейронной сети¶
In [19]:
np.random.seed(random_state)
tf.random.set_seed(random_state)
features_df = pd.read_csv('features.csv')
X = features_df.drop(columns = ["Класс"])
y = features_df["Класс"].values
y = to_categorical(y)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=random_state)
scaler = MinMaxScaler()
X_train = scaler.fit_transform(X_train)
X_test = scaler.transform(X_test)
model = Sequential()
model.add(Dense(128, input_shape=(186,), activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(512, activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(256, activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(64, activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(5, activation='softmax'))
early_stopping = EarlyStopping(monitor='val_loss', patience=200, verbose=1, restore_best_weights=True)
model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])
trained_model = model.fit(X_train, y_train, epochs=1000, batch_size=128, validation_split=0.2, callbacks=[early_stopping])
y_pred = model.predict(X_test, verbose=0)
y_pred_classes = np.argmax(y_pred, axis=1)
y_true = np.argmax(y_test, axis=1)
Epoch 1/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 2s 6ms/step - accuracy: 0.4889 - loss: 1.1808 - val_accuracy: 0.8066 - val_loss: 0.4973 Epoch 2/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.7538 - loss: 0.5807 - val_accuracy: 0.8320 - val_loss: 0.4059 Epoch 3/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.7911 - loss: 0.4908 - val_accuracy: 0.8456 - val_loss: 0.3838 Epoch 4/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8103 - loss: 0.4466 - val_accuracy: 0.8546 - val_loss: 0.3556 Epoch 5/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8165 - loss: 0.4313 - val_accuracy: 0.8501 - val_loss: 0.3522 Epoch 6/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8297 - loss: 0.4013 - val_accuracy: 0.8379 - val_loss: 0.3868 Epoch 7/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8363 - loss: 0.3996 - val_accuracy: 0.8546 - val_loss: 0.3420 Epoch 8/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8283 - loss: 0.4052 - val_accuracy: 0.8578 - val_loss: 0.3369 Epoch 9/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8462 - loss: 0.3759 - val_accuracy: 0.8519 - val_loss: 0.3306 Epoch 10/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8292 - loss: 0.3982 - val_accuracy: 0.8582 - val_loss: 0.3271 Epoch 11/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8422 - loss: 0.3698 - val_accuracy: 0.8519 - val_loss: 0.3462 Epoch 12/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8409 - loss: 0.3678 - val_accuracy: 0.8619 - val_loss: 0.3159 Epoch 13/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8475 - loss: 0.3573 - val_accuracy: 0.8601 - val_loss: 0.3254 Epoch 14/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8351 - loss: 0.3835 - val_accuracy: 0.8510 - val_loss: 0.3396 Epoch 15/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8497 - loss: 0.3493 - val_accuracy: 0.8655 - val_loss: 0.3143 Epoch 16/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8560 - loss: 0.3434 - val_accuracy: 0.8623 - val_loss: 0.3251 Epoch 17/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8517 - loss: 0.3466 - val_accuracy: 0.8605 - val_loss: 0.3111 Epoch 18/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8530 - loss: 0.3484 - val_accuracy: 0.8605 - val_loss: 0.3197 Epoch 19/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8551 - loss: 0.3439 - val_accuracy: 0.8619 - val_loss: 0.3103 Epoch 20/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8531 - loss: 0.3401 - val_accuracy: 0.8537 - val_loss: 0.3445 Epoch 21/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8528 - loss: 0.3550 - val_accuracy: 0.8551 - val_loss: 0.3165 Epoch 22/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8507 - loss: 0.3432 - val_accuracy: 0.8601 - val_loss: 0.3168 Epoch 23/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8506 - loss: 0.3519 - val_accuracy: 0.8623 - val_loss: 0.3203 Epoch 24/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8563 - loss: 0.3413 - val_accuracy: 0.8537 - val_loss: 0.3479 Epoch 25/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8622 - loss: 0.3258 - val_accuracy: 0.8673 - val_loss: 0.3053 Epoch 26/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8533 - loss: 0.3359 - val_accuracy: 0.8714 - val_loss: 0.3067 Epoch 27/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8628 - loss: 0.3278 - val_accuracy: 0.8700 - val_loss: 0.3050 Epoch 28/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8675 - loss: 0.3193 - val_accuracy: 0.8700 - val_loss: 0.3072 Epoch 29/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8659 - loss: 0.3265 - val_accuracy: 0.8546 - val_loss: 0.3513 Epoch 30/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8573 - loss: 0.3308 - val_accuracy: 0.8623 - val_loss: 0.3230 Epoch 31/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8687 - loss: 0.3204 - val_accuracy: 0.8587 - val_loss: 0.3321 Epoch 32/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8648 - loss: 0.3340 - val_accuracy: 0.8736 - val_loss: 0.2982 Epoch 33/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8667 - loss: 0.3252 - val_accuracy: 0.8569 - val_loss: 0.3348 Epoch 34/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8659 - loss: 0.3162 - val_accuracy: 0.8759 - val_loss: 0.3021 Epoch 35/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8771 - loss: 0.3081 - val_accuracy: 0.8714 - val_loss: 0.3094 Epoch 36/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8673 - loss: 0.3106 - val_accuracy: 0.8610 - val_loss: 0.3373 Epoch 37/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8690 - loss: 0.3098 - val_accuracy: 0.8705 - val_loss: 0.3167 Epoch 38/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8624 - loss: 0.3225 - val_accuracy: 0.8605 - val_loss: 0.3218 Epoch 39/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8729 - loss: 0.3067 - val_accuracy: 0.8659 - val_loss: 0.3086 Epoch 40/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8718 - loss: 0.3057 - val_accuracy: 0.8623 - val_loss: 0.3138 Epoch 41/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8726 - loss: 0.3004 - val_accuracy: 0.8777 - val_loss: 0.2893 Epoch 42/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8697 - loss: 0.3123 - val_accuracy: 0.8741 - val_loss: 0.3055 Epoch 43/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8726 - loss: 0.3106 - val_accuracy: 0.8678 - val_loss: 0.2934 Epoch 44/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8668 - loss: 0.3100 - val_accuracy: 0.8591 - val_loss: 0.3417 Epoch 45/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8714 - loss: 0.3061 - val_accuracy: 0.8696 - val_loss: 0.2955 Epoch 46/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8715 - loss: 0.2977 - val_accuracy: 0.8632 - val_loss: 0.3398 Epoch 47/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8760 - loss: 0.2994 - val_accuracy: 0.8619 - val_loss: 0.3272 Epoch 48/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8708 - loss: 0.3018 - val_accuracy: 0.8705 - val_loss: 0.3051 Epoch 49/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8750 - loss: 0.2983 - val_accuracy: 0.8682 - val_loss: 0.3036 Epoch 50/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8765 - loss: 0.2953 - val_accuracy: 0.8718 - val_loss: 0.3019 Epoch 51/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8808 - loss: 0.2902 - val_accuracy: 0.8564 - val_loss: 0.3412 Epoch 52/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8721 - loss: 0.3054 - val_accuracy: 0.8777 - val_loss: 0.3056 Epoch 53/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8766 - loss: 0.2913 - val_accuracy: 0.8687 - val_loss: 0.2964 Epoch 54/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8803 - loss: 0.2966 - val_accuracy: 0.8795 - val_loss: 0.2856 Epoch 55/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.8779 - loss: 0.2948 - val_accuracy: 0.8777 - val_loss: 0.3007 Epoch 56/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8752 - loss: 0.2968 - val_accuracy: 0.8863 - val_loss: 0.2859 Epoch 57/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8785 - loss: 0.2925 - val_accuracy: 0.8732 - val_loss: 0.2889 Epoch 58/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8736 - loss: 0.2979 - val_accuracy: 0.8741 - val_loss: 0.3049 Epoch 59/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8776 - loss: 0.2881 - val_accuracy: 0.8678 - val_loss: 0.3408 Epoch 60/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8806 - loss: 0.2804 - val_accuracy: 0.8723 - val_loss: 0.2967 Epoch 61/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8791 - loss: 0.2882 - val_accuracy: 0.8813 - val_loss: 0.2957 Epoch 62/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8797 - loss: 0.2806 - val_accuracy: 0.8723 - val_loss: 0.3031 Epoch 63/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8852 - loss: 0.2779 - val_accuracy: 0.8691 - val_loss: 0.3387 Epoch 64/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8767 - loss: 0.2861 - val_accuracy: 0.8483 - val_loss: 0.4013 Epoch 65/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8769 - loss: 0.2936 - val_accuracy: 0.8664 - val_loss: 0.3355 Epoch 66/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8779 - loss: 0.2856 - val_accuracy: 0.8759 - val_loss: 0.3049 Epoch 67/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8796 - loss: 0.2779 - val_accuracy: 0.8764 - val_loss: 0.2857 Epoch 68/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8827 - loss: 0.2914 - val_accuracy: 0.8755 - val_loss: 0.2906 Epoch 69/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8800 - loss: 0.2799 - val_accuracy: 0.8750 - val_loss: 0.2853 Epoch 70/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8830 - loss: 0.2718 - val_accuracy: 0.8705 - val_loss: 0.3065 Epoch 71/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8822 - loss: 0.2814 - val_accuracy: 0.8723 - val_loss: 0.3031 Epoch 72/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8840 - loss: 0.2685 - val_accuracy: 0.8791 - val_loss: 0.2806 Epoch 73/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8802 - loss: 0.2789 - val_accuracy: 0.8773 - val_loss: 0.2985 Epoch 74/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8821 - loss: 0.2761 - val_accuracy: 0.8804 - val_loss: 0.2881 Epoch 75/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8783 - loss: 0.2835 - val_accuracy: 0.8822 - val_loss: 0.2868 Epoch 76/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8835 - loss: 0.2809 - val_accuracy: 0.8687 - val_loss: 0.3342 Epoch 77/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8822 - loss: 0.2784 - val_accuracy: 0.8755 - val_loss: 0.3038 Epoch 78/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8841 - loss: 0.2752 - val_accuracy: 0.8727 - val_loss: 0.3198 Epoch 79/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8792 - loss: 0.2807 - val_accuracy: 0.8755 - val_loss: 0.3011 Epoch 80/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8863 - loss: 0.2695 - val_accuracy: 0.8696 - val_loss: 0.3265 Epoch 81/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8802 - loss: 0.2752 - val_accuracy: 0.8836 - val_loss: 0.2855 Epoch 82/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8852 - loss: 0.2710 - val_accuracy: 0.8813 - val_loss: 0.2846 Epoch 83/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8890 - loss: 0.2719 - val_accuracy: 0.8813 - val_loss: 0.2967 Epoch 84/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.8862 - loss: 0.2807 - val_accuracy: 0.8786 - val_loss: 0.3001 Epoch 85/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.8868 - loss: 0.2710 - val_accuracy: 0.8759 - val_loss: 0.2911 Epoch 86/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8765 - loss: 0.2857 - val_accuracy: 0.8813 - val_loss: 0.2923 Epoch 87/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8833 - loss: 0.2737 - val_accuracy: 0.8827 - val_loss: 0.2971 Epoch 88/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8848 - loss: 0.2751 - val_accuracy: 0.8859 - val_loss: 0.2705 Epoch 89/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8843 - loss: 0.2697 - val_accuracy: 0.8714 - val_loss: 0.3201 Epoch 90/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8868 - loss: 0.2783 - val_accuracy: 0.8854 - val_loss: 0.2806 Epoch 91/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8889 - loss: 0.2675 - val_accuracy: 0.8650 - val_loss: 0.3145 Epoch 92/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8812 - loss: 0.2732 - val_accuracy: 0.8782 - val_loss: 0.3104 Epoch 93/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8869 - loss: 0.2628 - val_accuracy: 0.8764 - val_loss: 0.2981 Epoch 94/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8840 - loss: 0.2787 - val_accuracy: 0.8936 - val_loss: 0.2652 Epoch 95/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8924 - loss: 0.2532 - val_accuracy: 0.8841 - val_loss: 0.2832 Epoch 96/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8954 - loss: 0.2521 - val_accuracy: 0.8818 - val_loss: 0.2795 Epoch 97/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8895 - loss: 0.2568 - val_accuracy: 0.8700 - val_loss: 0.3461 Epoch 98/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8886 - loss: 0.2678 - val_accuracy: 0.8809 - val_loss: 0.2891 Epoch 99/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8911 - loss: 0.2641 - val_accuracy: 0.8764 - val_loss: 0.3056 Epoch 100/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8864 - loss: 0.2631 - val_accuracy: 0.8895 - val_loss: 0.2715 Epoch 101/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8938 - loss: 0.2483 - val_accuracy: 0.8804 - val_loss: 0.3002 Epoch 102/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8898 - loss: 0.2639 - val_accuracy: 0.8813 - val_loss: 0.2922 Epoch 103/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8896 - loss: 0.2632 - val_accuracy: 0.8777 - val_loss: 0.3127 Epoch 104/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8893 - loss: 0.2569 - val_accuracy: 0.8899 - val_loss: 0.2594 Epoch 105/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8893 - loss: 0.2607 - val_accuracy: 0.8868 - val_loss: 0.2907 Epoch 106/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8935 - loss: 0.2493 - val_accuracy: 0.8890 - val_loss: 0.2627 Epoch 107/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8929 - loss: 0.2506 - val_accuracy: 0.8723 - val_loss: 0.3380 Epoch 108/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.8951 - loss: 0.2515 - val_accuracy: 0.8773 - val_loss: 0.2847 Epoch 109/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8947 - loss: 0.2486 - val_accuracy: 0.8841 - val_loss: 0.2734 Epoch 110/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8894 - loss: 0.2629 - val_accuracy: 0.8931 - val_loss: 0.2636 Epoch 111/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8948 - loss: 0.2529 - val_accuracy: 0.8922 - val_loss: 0.2563 Epoch 112/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8951 - loss: 0.2494 - val_accuracy: 0.8836 - val_loss: 0.2883 Epoch 113/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8963 - loss: 0.2514 - val_accuracy: 0.8786 - val_loss: 0.3075 Epoch 114/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8894 - loss: 0.2720 - val_accuracy: 0.8827 - val_loss: 0.2969 Epoch 115/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8926 - loss: 0.2478 - val_accuracy: 0.8909 - val_loss: 0.2601 Epoch 116/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8973 - loss: 0.2424 - val_accuracy: 0.8927 - val_loss: 0.2697 Epoch 117/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8998 - loss: 0.2444 - val_accuracy: 0.8886 - val_loss: 0.2677 Epoch 118/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8938 - loss: 0.2524 - val_accuracy: 0.8841 - val_loss: 0.2815 Epoch 119/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8983 - loss: 0.2328 - val_accuracy: 0.8872 - val_loss: 0.2846 Epoch 120/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8976 - loss: 0.2466 - val_accuracy: 0.8918 - val_loss: 0.2648 Epoch 121/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8978 - loss: 0.2460 - val_accuracy: 0.8832 - val_loss: 0.2821 Epoch 122/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8965 - loss: 0.2458 - val_accuracy: 0.8881 - val_loss: 0.2591 Epoch 123/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8967 - loss: 0.2501 - val_accuracy: 0.8868 - val_loss: 0.2712 Epoch 124/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8924 - loss: 0.2497 - val_accuracy: 0.8881 - val_loss: 0.2880 Epoch 125/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8930 - loss: 0.2473 - val_accuracy: 0.8800 - val_loss: 0.2925 Epoch 126/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8965 - loss: 0.2401 - val_accuracy: 0.8895 - val_loss: 0.2678 Epoch 127/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9001 - loss: 0.2373 - val_accuracy: 0.8886 - val_loss: 0.2725 Epoch 128/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9036 - loss: 0.2346 - val_accuracy: 0.8822 - val_loss: 0.2867 Epoch 129/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9036 - loss: 0.2323 - val_accuracy: 0.8872 - val_loss: 0.2782 Epoch 130/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9043 - loss: 0.2336 - val_accuracy: 0.8818 - val_loss: 0.2938 Epoch 131/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8999 - loss: 0.2375 - val_accuracy: 0.8795 - val_loss: 0.2967 Epoch 132/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9001 - loss: 0.2388 - val_accuracy: 0.8904 - val_loss: 0.2709 Epoch 133/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9032 - loss: 0.2276 - val_accuracy: 0.8764 - val_loss: 0.3278 Epoch 134/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9018 - loss: 0.2336 - val_accuracy: 0.8745 - val_loss: 0.3106 Epoch 135/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8996 - loss: 0.2388 - val_accuracy: 0.8813 - val_loss: 0.2886 Epoch 136/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8994 - loss: 0.2384 - val_accuracy: 0.8958 - val_loss: 0.2546 Epoch 137/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9040 - loss: 0.2279 - val_accuracy: 0.8818 - val_loss: 0.2886 Epoch 138/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8998 - loss: 0.2398 - val_accuracy: 0.8773 - val_loss: 0.3024 Epoch 139/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8965 - loss: 0.2452 - val_accuracy: 0.8895 - val_loss: 0.2718 Epoch 140/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9044 - loss: 0.2298 - val_accuracy: 0.8972 - val_loss: 0.2709 Epoch 141/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9069 - loss: 0.2228 - val_accuracy: 0.8868 - val_loss: 0.2912 Epoch 142/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8976 - loss: 0.2475 - val_accuracy: 0.8836 - val_loss: 0.2886 Epoch 143/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9007 - loss: 0.2432 - val_accuracy: 0.8863 - val_loss: 0.2851 Epoch 144/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9028 - loss: 0.2325 - val_accuracy: 0.8904 - val_loss: 0.2741 Epoch 145/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9058 - loss: 0.2351 - val_accuracy: 0.8967 - val_loss: 0.2563 Epoch 146/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9096 - loss: 0.2319 - val_accuracy: 0.8986 - val_loss: 0.2514 Epoch 147/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9000 - loss: 0.2416 - val_accuracy: 0.8863 - val_loss: 0.2879 Epoch 148/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9019 - loss: 0.2352 - val_accuracy: 0.8967 - val_loss: 0.2644 Epoch 149/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9055 - loss: 0.2287 - val_accuracy: 0.8836 - val_loss: 0.3212 Epoch 150/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9022 - loss: 0.2296 - val_accuracy: 0.8909 - val_loss: 0.2861 Epoch 151/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9078 - loss: 0.2209 - val_accuracy: 0.8981 - val_loss: 0.2499 Epoch 152/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9067 - loss: 0.2190 - val_accuracy: 0.9013 - val_loss: 0.2451 Epoch 153/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9131 - loss: 0.2174 - val_accuracy: 0.8976 - val_loss: 0.2595 Epoch 154/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.8999 - loss: 0.2294 - val_accuracy: 0.8863 - val_loss: 0.2768 Epoch 155/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9074 - loss: 0.2143 - val_accuracy: 0.8972 - val_loss: 0.2679 Epoch 156/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9062 - loss: 0.2262 - val_accuracy: 0.8904 - val_loss: 0.2731 Epoch 157/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9076 - loss: 0.2218 - val_accuracy: 0.8940 - val_loss: 0.2918 Epoch 158/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9082 - loss: 0.2196 - val_accuracy: 0.8899 - val_loss: 0.2954 Epoch 159/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9097 - loss: 0.2222 - val_accuracy: 0.8700 - val_loss: 0.3434 Epoch 160/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9103 - loss: 0.2221 - val_accuracy: 0.8972 - val_loss: 0.2579 Epoch 161/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9081 - loss: 0.2125 - val_accuracy: 0.8954 - val_loss: 0.2810 Epoch 162/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9105 - loss: 0.2165 - val_accuracy: 0.8705 - val_loss: 0.3455 Epoch 163/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9041 - loss: 0.2370 - val_accuracy: 0.8909 - val_loss: 0.2815 Epoch 164/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9104 - loss: 0.2190 - val_accuracy: 0.8909 - val_loss: 0.2763 Epoch 165/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9085 - loss: 0.2211 - val_accuracy: 0.8895 - val_loss: 0.2752 Epoch 166/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9120 - loss: 0.2157 - val_accuracy: 0.8909 - val_loss: 0.2934 Epoch 167/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9118 - loss: 0.2120 - val_accuracy: 0.8922 - val_loss: 0.2919 Epoch 168/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9080 - loss: 0.2229 - val_accuracy: 0.8999 - val_loss: 0.2602 Epoch 169/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9198 - loss: 0.2056 - val_accuracy: 0.8945 - val_loss: 0.2795 Epoch 170/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9197 - loss: 0.2009 - val_accuracy: 0.8863 - val_loss: 0.3091 Epoch 171/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9143 - loss: 0.2139 - val_accuracy: 0.9035 - val_loss: 0.2605 Epoch 172/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9123 - loss: 0.2094 - val_accuracy: 0.9004 - val_loss: 0.2616 Epoch 173/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9170 - loss: 0.2018 - val_accuracy: 0.8995 - val_loss: 0.2727 Epoch 174/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9138 - loss: 0.2112 - val_accuracy: 0.8981 - val_loss: 0.2667 Epoch 175/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9194 - loss: 0.2015 - val_accuracy: 0.9004 - val_loss: 0.2897 Epoch 176/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9192 - loss: 0.2006 - val_accuracy: 0.8972 - val_loss: 0.2951 Epoch 177/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9118 - loss: 0.2047 - val_accuracy: 0.9013 - val_loss: 0.2755 Epoch 178/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9166 - loss: 0.1985 - val_accuracy: 0.8786 - val_loss: 0.3701 Epoch 179/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9119 - loss: 0.2090 - val_accuracy: 0.9081 - val_loss: 0.2494 Epoch 180/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9174 - loss: 0.2023 - val_accuracy: 0.8927 - val_loss: 0.2955 Epoch 181/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9177 - loss: 0.2049 - val_accuracy: 0.8918 - val_loss: 0.3023 Epoch 182/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9224 - loss: 0.1926 - val_accuracy: 0.8895 - val_loss: 0.3134 Epoch 183/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9166 - loss: 0.2085 - val_accuracy: 0.9085 - val_loss: 0.2384 Epoch 184/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9186 - loss: 0.2010 - val_accuracy: 0.9017 - val_loss: 0.2595 Epoch 185/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9210 - loss: 0.2032 - val_accuracy: 0.8904 - val_loss: 0.2814 Epoch 186/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9217 - loss: 0.2065 - val_accuracy: 0.9099 - val_loss: 0.2655 Epoch 187/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9271 - loss: 0.1905 - val_accuracy: 0.9026 - val_loss: 0.2687 Epoch 188/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9219 - loss: 0.1936 - val_accuracy: 0.8809 - val_loss: 0.3246 Epoch 189/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9150 - loss: 0.1986 - val_accuracy: 0.9103 - val_loss: 0.2446 Epoch 190/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9202 - loss: 0.1969 - val_accuracy: 0.8922 - val_loss: 0.3071 Epoch 191/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9209 - loss: 0.2020 - val_accuracy: 0.9017 - val_loss: 0.2608 Epoch 192/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9257 - loss: 0.1872 - val_accuracy: 0.8909 - val_loss: 0.2979 Epoch 193/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9221 - loss: 0.1924 - val_accuracy: 0.8881 - val_loss: 0.2921 Epoch 194/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9165 - loss: 0.2031 - val_accuracy: 0.9085 - val_loss: 0.2643 Epoch 195/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9225 - loss: 0.1938 - val_accuracy: 0.9044 - val_loss: 0.2558 Epoch 196/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9193 - loss: 0.1948 - val_accuracy: 0.8981 - val_loss: 0.2697 Epoch 197/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9209 - loss: 0.1951 - val_accuracy: 0.9026 - val_loss: 0.2755 Epoch 198/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9108 - loss: 0.2163 - val_accuracy: 0.9031 - val_loss: 0.2628 Epoch 199/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9264 - loss: 0.1852 - val_accuracy: 0.9072 - val_loss: 0.2471 Epoch 200/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9222 - loss: 0.1885 - val_accuracy: 0.9130 - val_loss: 0.2581 Epoch 201/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9211 - loss: 0.2095 - val_accuracy: 0.9008 - val_loss: 0.2876 Epoch 202/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9290 - loss: 0.1829 - val_accuracy: 0.8990 - val_loss: 0.3002 Epoch 203/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9203 - loss: 0.1910 - val_accuracy: 0.8895 - val_loss: 0.3043 Epoch 204/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9319 - loss: 0.1773 - val_accuracy: 0.9035 - val_loss: 0.2774 Epoch 205/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9264 - loss: 0.1840 - val_accuracy: 0.8931 - val_loss: 0.2965 Epoch 206/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9187 - loss: 0.1956 - val_accuracy: 0.8986 - val_loss: 0.3133 Epoch 207/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9327 - loss: 0.1728 - val_accuracy: 0.9090 - val_loss: 0.2765 Epoch 208/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9252 - loss: 0.1865 - val_accuracy: 0.9026 - val_loss: 0.2685 Epoch 209/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9261 - loss: 0.1854 - val_accuracy: 0.9044 - val_loss: 0.2768 Epoch 210/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9286 - loss: 0.1886 - val_accuracy: 0.9049 - val_loss: 0.2576 Epoch 211/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9196 - loss: 0.2042 - val_accuracy: 0.9044 - val_loss: 0.2694 Epoch 212/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9283 - loss: 0.1873 - val_accuracy: 0.9040 - val_loss: 0.2799 Epoch 213/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9215 - loss: 0.1992 - val_accuracy: 0.8881 - val_loss: 0.3211 Epoch 214/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9210 - loss: 0.1952 - val_accuracy: 0.9117 - val_loss: 0.2482 Epoch 215/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9239 - loss: 0.1872 - val_accuracy: 0.8981 - val_loss: 0.3241 Epoch 216/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9306 - loss: 0.1835 - val_accuracy: 0.9049 - val_loss: 0.2919 Epoch 217/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9253 - loss: 0.1914 - val_accuracy: 0.9153 - val_loss: 0.2421 Epoch 218/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9306 - loss: 0.1787 - val_accuracy: 0.9108 - val_loss: 0.2607 Epoch 219/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9314 - loss: 0.1812 - val_accuracy: 0.9067 - val_loss: 0.2820 Epoch 220/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9258 - loss: 0.1948 - val_accuracy: 0.9094 - val_loss: 0.2566 Epoch 221/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9318 - loss: 0.1750 - val_accuracy: 0.8986 - val_loss: 0.2932 Epoch 222/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9259 - loss: 0.1842 - val_accuracy: 0.9126 - val_loss: 0.2337 Epoch 223/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9293 - loss: 0.1771 - val_accuracy: 0.9035 - val_loss: 0.2675 Epoch 224/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9296 - loss: 0.1739 - val_accuracy: 0.9126 - val_loss: 0.2448 Epoch 225/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9296 - loss: 0.1848 - val_accuracy: 0.8999 - val_loss: 0.2840 Epoch 226/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9217 - loss: 0.1955 - val_accuracy: 0.9022 - val_loss: 0.2816 Epoch 227/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9290 - loss: 0.1821 - val_accuracy: 0.9117 - val_loss: 0.2435 Epoch 228/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9278 - loss: 0.1833 - val_accuracy: 0.9135 - val_loss: 0.2594 Epoch 229/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9256 - loss: 0.1911 - val_accuracy: 0.9099 - val_loss: 0.2691 Epoch 230/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9333 - loss: 0.1698 - val_accuracy: 0.9153 - val_loss: 0.2545 Epoch 231/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9351 - loss: 0.1680 - val_accuracy: 0.9072 - val_loss: 0.2872 Epoch 232/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9316 - loss: 0.1790 - val_accuracy: 0.9135 - val_loss: 0.2557 Epoch 233/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9294 - loss: 0.1773 - val_accuracy: 0.9022 - val_loss: 0.2830 Epoch 234/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9259 - loss: 0.1839 - val_accuracy: 0.9194 - val_loss: 0.2252 Epoch 235/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9393 - loss: 0.1605 - val_accuracy: 0.8940 - val_loss: 0.3329 Epoch 236/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9299 - loss: 0.1846 - val_accuracy: 0.8918 - val_loss: 0.3170 Epoch 237/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9350 - loss: 0.1750 - val_accuracy: 0.9158 - val_loss: 0.2404 Epoch 238/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9298 - loss: 0.1721 - val_accuracy: 0.9062 - val_loss: 0.2729 Epoch 239/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9291 - loss: 0.1792 - val_accuracy: 0.9094 - val_loss: 0.2694 Epoch 240/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9325 - loss: 0.1647 - val_accuracy: 0.9031 - val_loss: 0.2724 Epoch 241/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9340 - loss: 0.1709 - val_accuracy: 0.9017 - val_loss: 0.2854 Epoch 242/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9332 - loss: 0.1714 - val_accuracy: 0.9130 - val_loss: 0.2560 Epoch 243/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9338 - loss: 0.1781 - val_accuracy: 0.8986 - val_loss: 0.3394 Epoch 244/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9348 - loss: 0.1664 - val_accuracy: 0.9198 - val_loss: 0.2439 Epoch 245/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9291 - loss: 0.1709 - val_accuracy: 0.9090 - val_loss: 0.2930 Epoch 246/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9328 - loss: 0.1748 - val_accuracy: 0.9094 - val_loss: 0.2757 Epoch 247/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9388 - loss: 0.1618 - val_accuracy: 0.9121 - val_loss: 0.2721 Epoch 248/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9330 - loss: 0.1748 - val_accuracy: 0.9149 - val_loss: 0.2410 Epoch 249/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9341 - loss: 0.1718 - val_accuracy: 0.9149 - val_loss: 0.2490 Epoch 250/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9321 - loss: 0.1767 - val_accuracy: 0.9185 - val_loss: 0.2501 Epoch 251/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9332 - loss: 0.1650 - val_accuracy: 0.9099 - val_loss: 0.2660 Epoch 252/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9341 - loss: 0.1702 - val_accuracy: 0.9094 - val_loss: 0.2606 Epoch 253/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9290 - loss: 0.1789 - val_accuracy: 0.9226 - val_loss: 0.2347 Epoch 254/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9405 - loss: 0.1572 - val_accuracy: 0.9144 - val_loss: 0.2610 Epoch 255/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9334 - loss: 0.1723 - val_accuracy: 0.9253 - val_loss: 0.2113 Epoch 256/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9397 - loss: 0.1662 - val_accuracy: 0.9153 - val_loss: 0.2534 Epoch 257/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9283 - loss: 0.1746 - val_accuracy: 0.9216 - val_loss: 0.2318 Epoch 258/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9363 - loss: 0.1693 - val_accuracy: 0.9099 - val_loss: 0.2612 Epoch 259/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9357 - loss: 0.1651 - val_accuracy: 0.9253 - val_loss: 0.2180 Epoch 260/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9350 - loss: 0.1666 - val_accuracy: 0.9203 - val_loss: 0.2696 Epoch 261/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9419 - loss: 0.1556 - val_accuracy: 0.9044 - val_loss: 0.2886 Epoch 262/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9346 - loss: 0.1741 - val_accuracy: 0.9171 - val_loss: 0.2508 Epoch 263/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9365 - loss: 0.1731 - val_accuracy: 0.9153 - val_loss: 0.2472 Epoch 264/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9346 - loss: 0.1592 - val_accuracy: 0.9176 - val_loss: 0.2592 Epoch 265/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9347 - loss: 0.1740 - val_accuracy: 0.9158 - val_loss: 0.2556 Epoch 266/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9401 - loss: 0.1577 - val_accuracy: 0.9207 - val_loss: 0.2430 Epoch 267/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9363 - loss: 0.1682 - val_accuracy: 0.9194 - val_loss: 0.2425 Epoch 268/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9290 - loss: 0.1778 - val_accuracy: 0.9262 - val_loss: 0.2285 Epoch 269/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9385 - loss: 0.1539 - val_accuracy: 0.9162 - val_loss: 0.2626 Epoch 270/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9388 - loss: 0.1591 - val_accuracy: 0.9194 - val_loss: 0.2361 Epoch 271/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9403 - loss: 0.1560 - val_accuracy: 0.9185 - val_loss: 0.2546 Epoch 272/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9406 - loss: 0.1481 - val_accuracy: 0.9126 - val_loss: 0.2480 Epoch 273/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9404 - loss: 0.1690 - val_accuracy: 0.9126 - val_loss: 0.2555 Epoch 274/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9382 - loss: 0.1550 - val_accuracy: 0.9022 - val_loss: 0.2786 Epoch 275/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9346 - loss: 0.1719 - val_accuracy: 0.9198 - val_loss: 0.2494 Epoch 276/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9327 - loss: 0.1682 - val_accuracy: 0.9144 - val_loss: 0.2585 Epoch 277/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9451 - loss: 0.1446 - val_accuracy: 0.9198 - val_loss: 0.2546 Epoch 278/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9487 - loss: 0.1404 - val_accuracy: 0.9130 - val_loss: 0.2721 Epoch 279/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9438 - loss: 0.1490 - val_accuracy: 0.9189 - val_loss: 0.2634 Epoch 280/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9400 - loss: 0.1525 - val_accuracy: 0.9244 - val_loss: 0.2428 Epoch 281/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9431 - loss: 0.1517 - val_accuracy: 0.9103 - val_loss: 0.2626 Epoch 282/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9397 - loss: 0.1573 - val_accuracy: 0.9212 - val_loss: 0.2508 Epoch 283/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9434 - loss: 0.1511 - val_accuracy: 0.9149 - val_loss: 0.2637 Epoch 284/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9380 - loss: 0.1603 - val_accuracy: 0.9248 - val_loss: 0.2275 Epoch 285/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9432 - loss: 0.1554 - val_accuracy: 0.9099 - val_loss: 0.2815 Epoch 286/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9409 - loss: 0.1524 - val_accuracy: 0.9271 - val_loss: 0.2278 Epoch 287/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9415 - loss: 0.1505 - val_accuracy: 0.9244 - val_loss: 0.2406 Epoch 288/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9436 - loss: 0.1468 - val_accuracy: 0.9171 - val_loss: 0.2575 Epoch 289/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9373 - loss: 0.1615 - val_accuracy: 0.9158 - val_loss: 0.2519 Epoch 290/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9349 - loss: 0.1601 - val_accuracy: 0.9194 - val_loss: 0.2367 Epoch 291/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9397 - loss: 0.1544 - val_accuracy: 0.9207 - val_loss: 0.2451 Epoch 292/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9470 - loss: 0.1405 - val_accuracy: 0.9212 - val_loss: 0.2307 Epoch 293/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9402 - loss: 0.1598 - val_accuracy: 0.9198 - val_loss: 0.2316 Epoch 294/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9443 - loss: 0.1445 - val_accuracy: 0.9198 - val_loss: 0.2739 Epoch 295/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9396 - loss: 0.1499 - val_accuracy: 0.9185 - val_loss: 0.2398 Epoch 296/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9346 - loss: 0.1584 - val_accuracy: 0.9257 - val_loss: 0.2471 Epoch 297/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9458 - loss: 0.1455 - val_accuracy: 0.9226 - val_loss: 0.2423 Epoch 298/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9456 - loss: 0.1441 - val_accuracy: 0.9248 - val_loss: 0.2520 Epoch 299/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9423 - loss: 0.1537 - val_accuracy: 0.9198 - val_loss: 0.2549 Epoch 300/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9390 - loss: 0.1586 - val_accuracy: 0.9275 - val_loss: 0.2269 Epoch 301/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9355 - loss: 0.1606 - val_accuracy: 0.9280 - val_loss: 0.2185 Epoch 302/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9429 - loss: 0.1448 - val_accuracy: 0.9284 - val_loss: 0.2337 Epoch 303/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9446 - loss: 0.1449 - val_accuracy: 0.9230 - val_loss: 0.2182 Epoch 304/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9429 - loss: 0.1501 - val_accuracy: 0.9366 - val_loss: 0.2012 Epoch 305/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9516 - loss: 0.1288 - val_accuracy: 0.9203 - val_loss: 0.2535 Epoch 306/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9463 - loss: 0.1428 - val_accuracy: 0.9275 - val_loss: 0.2294 Epoch 307/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9439 - loss: 0.1510 - val_accuracy: 0.9171 - val_loss: 0.2594 Epoch 308/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9427 - loss: 0.1433 - val_accuracy: 0.9207 - val_loss: 0.2438 Epoch 309/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9500 - loss: 0.1411 - val_accuracy: 0.9235 - val_loss: 0.2253 Epoch 310/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9482 - loss: 0.1332 - val_accuracy: 0.9099 - val_loss: 0.2599 Epoch 311/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9396 - loss: 0.1600 - val_accuracy: 0.9176 - val_loss: 0.2644 Epoch 312/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9512 - loss: 0.1360 - val_accuracy: 0.9244 - val_loss: 0.2188 Epoch 313/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9528 - loss: 0.1316 - val_accuracy: 0.9321 - val_loss: 0.2084 Epoch 314/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9443 - loss: 0.1463 - val_accuracy: 0.9216 - val_loss: 0.2342 Epoch 315/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9472 - loss: 0.1383 - val_accuracy: 0.9348 - val_loss: 0.2212 Epoch 316/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9428 - loss: 0.1488 - val_accuracy: 0.9235 - val_loss: 0.2669 Epoch 317/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9462 - loss: 0.1364 - val_accuracy: 0.9343 - val_loss: 0.2065 Epoch 318/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9480 - loss: 0.1346 - val_accuracy: 0.9244 - val_loss: 0.2226 Epoch 319/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9478 - loss: 0.1349 - val_accuracy: 0.9235 - val_loss: 0.2305 Epoch 320/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9416 - loss: 0.1504 - val_accuracy: 0.9289 - val_loss: 0.2268 Epoch 321/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9466 - loss: 0.1419 - val_accuracy: 0.9198 - val_loss: 0.2333 Epoch 322/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9478 - loss: 0.1320 - val_accuracy: 0.9235 - val_loss: 0.2293 Epoch 323/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9418 - loss: 0.1489 - val_accuracy: 0.9198 - val_loss: 0.2272 Epoch 324/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9402 - loss: 0.1518 - val_accuracy: 0.9293 - val_loss: 0.2267 Epoch 325/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9443 - loss: 0.1384 - val_accuracy: 0.9271 - val_loss: 0.2219 Epoch 326/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9512 - loss: 0.1242 - val_accuracy: 0.9321 - val_loss: 0.2219 Epoch 327/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9506 - loss: 0.1358 - val_accuracy: 0.9257 - val_loss: 0.2260 Epoch 328/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9466 - loss: 0.1423 - val_accuracy: 0.9226 - val_loss: 0.2449 Epoch 329/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9459 - loss: 0.1473 - val_accuracy: 0.9289 - val_loss: 0.2400 Epoch 330/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9474 - loss: 0.1361 - val_accuracy: 0.9321 - val_loss: 0.2215 Epoch 331/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9479 - loss: 0.1361 - val_accuracy: 0.9289 - val_loss: 0.2182 Epoch 332/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9495 - loss: 0.1284 - val_accuracy: 0.9293 - val_loss: 0.2285 Epoch 333/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9467 - loss: 0.1433 - val_accuracy: 0.9189 - val_loss: 0.2651 Epoch 334/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9475 - loss: 0.1381 - val_accuracy: 0.9307 - val_loss: 0.2235 Epoch 335/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9507 - loss: 0.1315 - val_accuracy: 0.9253 - val_loss: 0.2369 Epoch 336/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9493 - loss: 0.1355 - val_accuracy: 0.9275 - val_loss: 0.2318 Epoch 337/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9467 - loss: 0.1370 - val_accuracy: 0.9271 - val_loss: 0.2284 Epoch 338/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9524 - loss: 0.1237 - val_accuracy: 0.9289 - val_loss: 0.2388 Epoch 339/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9525 - loss: 0.1230 - val_accuracy: 0.9221 - val_loss: 0.2327 Epoch 340/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9505 - loss: 0.1471 - val_accuracy: 0.9303 - val_loss: 0.2314 Epoch 341/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9522 - loss: 0.1290 - val_accuracy: 0.9221 - val_loss: 0.2452 Epoch 342/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9535 - loss: 0.1239 - val_accuracy: 0.9248 - val_loss: 0.2446 Epoch 343/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9455 - loss: 0.1432 - val_accuracy: 0.9171 - val_loss: 0.2421 Epoch 344/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9470 - loss: 0.1325 - val_accuracy: 0.9189 - val_loss: 0.2561 Epoch 345/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9439 - loss: 0.1439 - val_accuracy: 0.9180 - val_loss: 0.2428 Epoch 346/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9506 - loss: 0.1368 - val_accuracy: 0.9289 - val_loss: 0.2273 Epoch 347/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9505 - loss: 0.1275 - val_accuracy: 0.9194 - val_loss: 0.2354 Epoch 348/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9565 - loss: 0.1184 - val_accuracy: 0.9275 - val_loss: 0.2354 Epoch 349/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9551 - loss: 0.1185 - val_accuracy: 0.9198 - val_loss: 0.2583 Epoch 350/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9543 - loss: 0.1255 - val_accuracy: 0.9235 - val_loss: 0.2255 Epoch 351/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9515 - loss: 0.1297 - val_accuracy: 0.9221 - val_loss: 0.2403 Epoch 352/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9432 - loss: 0.1520 - val_accuracy: 0.9253 - val_loss: 0.2137 Epoch 353/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9428 - loss: 0.1379 - val_accuracy: 0.9271 - val_loss: 0.2327 Epoch 354/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9550 - loss: 0.1232 - val_accuracy: 0.9275 - val_loss: 0.2228 Epoch 355/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9474 - loss: 0.1291 - val_accuracy: 0.9325 - val_loss: 0.2243 Epoch 356/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9555 - loss: 0.1279 - val_accuracy: 0.9298 - val_loss: 0.2300 Epoch 357/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9558 - loss: 0.1180 - val_accuracy: 0.9284 - val_loss: 0.2259 Epoch 358/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9558 - loss: 0.1177 - val_accuracy: 0.9303 - val_loss: 0.2254 Epoch 359/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9518 - loss: 0.1320 - val_accuracy: 0.9284 - val_loss: 0.2318 Epoch 360/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9523 - loss: 0.1250 - val_accuracy: 0.9239 - val_loss: 0.2523 Epoch 361/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9559 - loss: 0.1199 - val_accuracy: 0.9244 - val_loss: 0.2494 Epoch 362/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9549 - loss: 0.1250 - val_accuracy: 0.9275 - val_loss: 0.2260 Epoch 363/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9493 - loss: 0.1404 - val_accuracy: 0.9244 - val_loss: 0.2378 Epoch 364/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9506 - loss: 0.1211 - val_accuracy: 0.9248 - val_loss: 0.2520 Epoch 365/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9516 - loss: 0.1345 - val_accuracy: 0.9312 - val_loss: 0.2304 Epoch 366/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9496 - loss: 0.1313 - val_accuracy: 0.9312 - val_loss: 0.2167 Epoch 367/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9567 - loss: 0.1158 - val_accuracy: 0.9271 - val_loss: 0.2332 Epoch 368/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9565 - loss: 0.1135 - val_accuracy: 0.9266 - val_loss: 0.2416 Epoch 369/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9486 - loss: 0.1327 - val_accuracy: 0.9230 - val_loss: 0.2563 Epoch 370/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9556 - loss: 0.1222 - val_accuracy: 0.9316 - val_loss: 0.2197 Epoch 371/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9537 - loss: 0.1220 - val_accuracy: 0.9307 - val_loss: 0.2189 Epoch 372/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9549 - loss: 0.1189 - val_accuracy: 0.9343 - val_loss: 0.2110 Epoch 373/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9565 - loss: 0.1160 - val_accuracy: 0.9248 - val_loss: 0.2633 Epoch 374/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9541 - loss: 0.1248 - val_accuracy: 0.9303 - val_loss: 0.2500 Epoch 375/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9454 - loss: 0.1471 - val_accuracy: 0.9298 - val_loss: 0.2229 Epoch 376/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9584 - loss: 0.1096 - val_accuracy: 0.9275 - val_loss: 0.2222 Epoch 377/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9564 - loss: 0.1122 - val_accuracy: 0.9289 - val_loss: 0.2304 Epoch 378/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9503 - loss: 0.1274 - val_accuracy: 0.9284 - val_loss: 0.2368 Epoch 379/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9485 - loss: 0.1346 - val_accuracy: 0.9212 - val_loss: 0.2529 Epoch 380/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9516 - loss: 0.1252 - val_accuracy: 0.9167 - val_loss: 0.2680 Epoch 381/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9576 - loss: 0.1199 - val_accuracy: 0.9090 - val_loss: 0.2766 Epoch 382/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9491 - loss: 0.1388 - val_accuracy: 0.9275 - val_loss: 0.2292 Epoch 383/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9526 - loss: 0.1242 - val_accuracy: 0.9235 - val_loss: 0.2423 Epoch 384/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9449 - loss: 0.1382 - val_accuracy: 0.9321 - val_loss: 0.2395 Epoch 385/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9520 - loss: 0.1433 - val_accuracy: 0.9330 - val_loss: 0.2314 Epoch 386/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9531 - loss: 0.1237 - val_accuracy: 0.9298 - val_loss: 0.2125 Epoch 387/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9551 - loss: 0.1207 - val_accuracy: 0.9370 - val_loss: 0.2094 Epoch 388/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9550 - loss: 0.1193 - val_accuracy: 0.9230 - val_loss: 0.2429 Epoch 389/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9571 - loss: 0.1174 - val_accuracy: 0.9230 - val_loss: 0.2319 Epoch 390/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9529 - loss: 0.1252 - val_accuracy: 0.9244 - val_loss: 0.2328 Epoch 391/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9601 - loss: 0.1082 - val_accuracy: 0.9262 - val_loss: 0.2303 Epoch 392/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9515 - loss: 0.1309 - val_accuracy: 0.9244 - val_loss: 0.2220 Epoch 393/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9589 - loss: 0.1136 - val_accuracy: 0.9266 - val_loss: 0.2393 Epoch 394/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9590 - loss: 0.1133 - val_accuracy: 0.9284 - val_loss: 0.2120 Epoch 395/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9556 - loss: 0.1184 - val_accuracy: 0.9280 - val_loss: 0.2259 Epoch 396/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9526 - loss: 0.1298 - val_accuracy: 0.9275 - val_loss: 0.2387 Epoch 397/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9554 - loss: 0.1168 - val_accuracy: 0.9289 - val_loss: 0.2316 Epoch 398/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9599 - loss: 0.1152 - val_accuracy: 0.9266 - val_loss: 0.2353 Epoch 399/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9565 - loss: 0.1160 - val_accuracy: 0.9280 - val_loss: 0.2322 Epoch 400/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9510 - loss: 0.1260 - val_accuracy: 0.9312 - val_loss: 0.2131 Epoch 401/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9532 - loss: 0.1283 - val_accuracy: 0.9289 - val_loss: 0.2377 Epoch 402/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9560 - loss: 0.1136 - val_accuracy: 0.9361 - val_loss: 0.2153 Epoch 403/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9574 - loss: 0.1140 - val_accuracy: 0.9284 - val_loss: 0.2295 Epoch 404/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9598 - loss: 0.1096 - val_accuracy: 0.9244 - val_loss: 0.2543 Epoch 405/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9502 - loss: 0.1278 - val_accuracy: 0.9330 - val_loss: 0.2189 Epoch 406/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9527 - loss: 0.1193 - val_accuracy: 0.9244 - val_loss: 0.2522 Epoch 407/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9476 - loss: 0.1390 - val_accuracy: 0.9235 - val_loss: 0.2548 Epoch 408/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9472 - loss: 0.1366 - val_accuracy: 0.9198 - val_loss: 0.2782 Epoch 409/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9528 - loss: 0.1233 - val_accuracy: 0.9289 - val_loss: 0.2511 Epoch 410/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9537 - loss: 0.1194 - val_accuracy: 0.9334 - val_loss: 0.2103 Epoch 411/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9555 - loss: 0.1196 - val_accuracy: 0.9325 - val_loss: 0.2255 Epoch 412/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9608 - loss: 0.1038 - val_accuracy: 0.9253 - val_loss: 0.2500 Epoch 413/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9569 - loss: 0.1162 - val_accuracy: 0.9257 - val_loss: 0.2241 Epoch 414/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9552 - loss: 0.1178 - val_accuracy: 0.9226 - val_loss: 0.2357 Epoch 415/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9606 - loss: 0.1152 - val_accuracy: 0.9198 - val_loss: 0.2394 Epoch 416/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9571 - loss: 0.1112 - val_accuracy: 0.9280 - val_loss: 0.2572 Epoch 417/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9535 - loss: 0.1199 - val_accuracy: 0.9307 - val_loss: 0.2434 Epoch 418/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9606 - loss: 0.1098 - val_accuracy: 0.9325 - val_loss: 0.2182 Epoch 419/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9560 - loss: 0.1195 - val_accuracy: 0.9330 - val_loss: 0.2106 Epoch 420/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9585 - loss: 0.1057 - val_accuracy: 0.9303 - val_loss: 0.2188 Epoch 421/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9545 - loss: 0.1205 - val_accuracy: 0.9316 - val_loss: 0.2371 Epoch 422/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9598 - loss: 0.1114 - val_accuracy: 0.9212 - val_loss: 0.2452 Epoch 423/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9579 - loss: 0.1119 - val_accuracy: 0.9207 - val_loss: 0.2810 Epoch 424/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9539 - loss: 0.1234 - val_accuracy: 0.9266 - val_loss: 0.2558 Epoch 425/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9686 - loss: 0.0992 - val_accuracy: 0.9271 - val_loss: 0.2385 Epoch 426/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9573 - loss: 0.1174 - val_accuracy: 0.9203 - val_loss: 0.2897 Epoch 427/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9549 - loss: 0.1191 - val_accuracy: 0.9266 - val_loss: 0.2320 Epoch 428/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9610 - loss: 0.1087 - val_accuracy: 0.9185 - val_loss: 0.2632 Epoch 429/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9582 - loss: 0.1112 - val_accuracy: 0.9226 - val_loss: 0.2628 Epoch 430/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9589 - loss: 0.1117 - val_accuracy: 0.9266 - val_loss: 0.2406 Epoch 431/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9639 - loss: 0.0976 - val_accuracy: 0.9271 - val_loss: 0.2313 Epoch 432/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9666 - loss: 0.0970 - val_accuracy: 0.9189 - val_loss: 0.2703 Epoch 433/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9522 - loss: 0.1273 - val_accuracy: 0.9303 - val_loss: 0.2378 Epoch 434/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9620 - loss: 0.1061 - val_accuracy: 0.9375 - val_loss: 0.2187 Epoch 435/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9538 - loss: 0.1292 - val_accuracy: 0.9312 - val_loss: 0.2341 Epoch 436/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9566 - loss: 0.1154 - val_accuracy: 0.9339 - val_loss: 0.2204 Epoch 437/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9528 - loss: 0.1175 - val_accuracy: 0.9312 - val_loss: 0.2363 Epoch 438/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9598 - loss: 0.1089 - val_accuracy: 0.9307 - val_loss: 0.2276 Epoch 439/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9640 - loss: 0.0968 - val_accuracy: 0.9253 - val_loss: 0.2705 Epoch 440/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9555 - loss: 0.1133 - val_accuracy: 0.9239 - val_loss: 0.2454 Epoch 441/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9582 - loss: 0.1084 - val_accuracy: 0.9248 - val_loss: 0.2460 Epoch 442/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9572 - loss: 0.1161 - val_accuracy: 0.9099 - val_loss: 0.3091 Epoch 443/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9547 - loss: 0.1204 - val_accuracy: 0.9271 - val_loss: 0.2396 Epoch 444/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9576 - loss: 0.1185 - val_accuracy: 0.9189 - val_loss: 0.3063 Epoch 445/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9534 - loss: 0.1276 - val_accuracy: 0.9257 - val_loss: 0.2414 Epoch 446/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9637 - loss: 0.0987 - val_accuracy: 0.9248 - val_loss: 0.2431 Epoch 447/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9633 - loss: 0.1003 - val_accuracy: 0.9330 - val_loss: 0.2330 Epoch 448/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9571 - loss: 0.1089 - val_accuracy: 0.9230 - val_loss: 0.2776 Epoch 449/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9637 - loss: 0.1016 - val_accuracy: 0.9262 - val_loss: 0.2658 Epoch 450/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9612 - loss: 0.1022 - val_accuracy: 0.9298 - val_loss: 0.2459 Epoch 451/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9623 - loss: 0.0996 - val_accuracy: 0.9266 - val_loss: 0.2673 Epoch 452/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9582 - loss: 0.1085 - val_accuracy: 0.9253 - val_loss: 0.2277 Epoch 453/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9581 - loss: 0.1063 - val_accuracy: 0.9280 - val_loss: 0.2441 Epoch 454/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9648 - loss: 0.0984 - val_accuracy: 0.9275 - val_loss: 0.2540 Epoch 455/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9610 - loss: 0.1043 - val_accuracy: 0.9348 - val_loss: 0.2182 Epoch 456/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9568 - loss: 0.1244 - val_accuracy: 0.9266 - val_loss: 0.2430 Epoch 457/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9637 - loss: 0.0983 - val_accuracy: 0.9275 - val_loss: 0.2384 Epoch 458/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9611 - loss: 0.1046 - val_accuracy: 0.9280 - val_loss: 0.2689 Epoch 459/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9627 - loss: 0.1073 - val_accuracy: 0.9266 - val_loss: 0.2505 Epoch 460/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9650 - loss: 0.0925 - val_accuracy: 0.9207 - val_loss: 0.2726 Epoch 461/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9559 - loss: 0.1150 - val_accuracy: 0.9343 - val_loss: 0.2237 Epoch 462/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9570 - loss: 0.1174 - val_accuracy: 0.9257 - val_loss: 0.2292 Epoch 463/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9595 - loss: 0.1075 - val_accuracy: 0.9307 - val_loss: 0.2267 Epoch 464/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9653 - loss: 0.0924 - val_accuracy: 0.9339 - val_loss: 0.2349 Epoch 465/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9612 - loss: 0.1061 - val_accuracy: 0.9262 - val_loss: 0.2563 Epoch 466/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9568 - loss: 0.1127 - val_accuracy: 0.9325 - val_loss: 0.2395 Epoch 467/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9624 - loss: 0.1039 - val_accuracy: 0.9239 - val_loss: 0.2784 Epoch 468/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9594 - loss: 0.1061 - val_accuracy: 0.9298 - val_loss: 0.2348 Epoch 469/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9627 - loss: 0.1073 - val_accuracy: 0.9239 - val_loss: 0.2640 Epoch 470/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9604 - loss: 0.1002 - val_accuracy: 0.9307 - val_loss: 0.2590 Epoch 471/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9623 - loss: 0.1042 - val_accuracy: 0.9130 - val_loss: 0.3128 Epoch 472/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9607 - loss: 0.1088 - val_accuracy: 0.9239 - val_loss: 0.2425 Epoch 473/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9623 - loss: 0.0977 - val_accuracy: 0.9325 - val_loss: 0.2400 Epoch 474/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9594 - loss: 0.1055 - val_accuracy: 0.9248 - val_loss: 0.2733 Epoch 475/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9630 - loss: 0.1033 - val_accuracy: 0.9257 - val_loss: 0.2900 Epoch 476/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9593 - loss: 0.1100 - val_accuracy: 0.9289 - val_loss: 0.2374 Epoch 477/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9615 - loss: 0.1002 - val_accuracy: 0.9266 - val_loss: 0.2489 Epoch 478/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9612 - loss: 0.1101 - val_accuracy: 0.9180 - val_loss: 0.2742 Epoch 479/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9598 - loss: 0.1092 - val_accuracy: 0.9235 - val_loss: 0.2588 Epoch 480/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9636 - loss: 0.0973 - val_accuracy: 0.9289 - val_loss: 0.2579 Epoch 481/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9627 - loss: 0.1041 - val_accuracy: 0.9244 - val_loss: 0.2531 Epoch 482/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9619 - loss: 0.1029 - val_accuracy: 0.9253 - val_loss: 0.2486 Epoch 483/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9654 - loss: 0.0970 - val_accuracy: 0.9303 - val_loss: 0.2311 Epoch 484/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9602 - loss: 0.0985 - val_accuracy: 0.9284 - val_loss: 0.2398 Epoch 485/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9628 - loss: 0.1070 - val_accuracy: 0.9262 - val_loss: 0.2482 Epoch 486/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9588 - loss: 0.1091 - val_accuracy: 0.9298 - val_loss: 0.2375 Epoch 487/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9604 - loss: 0.1132 - val_accuracy: 0.9284 - val_loss: 0.2504 Epoch 488/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9633 - loss: 0.1029 - val_accuracy: 0.9253 - val_loss: 0.2590 Epoch 489/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9647 - loss: 0.0996 - val_accuracy: 0.9244 - val_loss: 0.2612 Epoch 490/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9687 - loss: 0.0917 - val_accuracy: 0.9230 - val_loss: 0.2707 Epoch 491/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9665 - loss: 0.0948 - val_accuracy: 0.9321 - val_loss: 0.2324 Epoch 492/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9641 - loss: 0.0912 - val_accuracy: 0.9321 - val_loss: 0.2493 Epoch 493/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9589 - loss: 0.1035 - val_accuracy: 0.9321 - val_loss: 0.2310 Epoch 494/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9660 - loss: 0.0982 - val_accuracy: 0.9271 - val_loss: 0.2571 Epoch 495/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9638 - loss: 0.0964 - val_accuracy: 0.9221 - val_loss: 0.2554 Epoch 496/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 5ms/step - accuracy: 0.9617 - loss: 0.1053 - val_accuracy: 0.9239 - val_loss: 0.2471 Epoch 497/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9624 - loss: 0.1012 - val_accuracy: 0.9293 - val_loss: 0.2669 Epoch 498/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9609 - loss: 0.1075 - val_accuracy: 0.9325 - val_loss: 0.2352 Epoch 499/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9562 - loss: 0.1125 - val_accuracy: 0.9235 - val_loss: 0.2679 Epoch 500/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9583 - loss: 0.1176 - val_accuracy: 0.9330 - val_loss: 0.2341 Epoch 501/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step - accuracy: 0.9660 - loss: 0.0948 - val_accuracy: 0.9303 - val_loss: 0.2449 Epoch 502/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9653 - loss: 0.0959 - val_accuracy: 0.9384 - val_loss: 0.2125 Epoch 503/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9636 - loss: 0.0936 - val_accuracy: 0.9262 - val_loss: 0.2664 Epoch 504/1000 69/69 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.9629 - loss: 0.1017 - val_accuracy: 0.9162 - val_loss: 0.2777 Epoch 504: early stopping Restoring model weights from the end of the best epoch: 304.
In [20]:
# Архитектура нейронной сети
model.summary()
Model: "sequential"
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┓ ┃ Layer (type) ┃ Output Shape ┃ Param # ┃ ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━┩ │ dense (Dense) │ (None, 128) │ 23,936 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ dropout (Dropout) │ (None, 128) │ 0 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ dense_1 (Dense) │ (None, 512) │ 66,048 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ dropout_1 (Dropout) │ (None, 512) │ 0 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ dense_2 (Dense) │ (None, 256) │ 131,328 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ dropout_2 (Dropout) │ (None, 256) │ 0 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ dense_3 (Dense) │ (None, 64) │ 16,448 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ dropout_3 (Dropout) │ (None, 64) │ 0 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ dense_4 (Dense) │ (None, 5) │ 325 │ └──────────────────────────────────────┴─────────────────────────────┴─────────────────┘
Total params: 714,257 (2.72 MB)
Trainable params: 238,085 (930.02 KB)
Non-trainable params: 0 (0.00 B)
Optimizer params: 476,172 (1.82 MB)
In [21]:
# Оценка нейронной сети по четырем метрикам точности
print(classification_report(y_true, y_pred_classes))
print(f" Accuracy: {round(accuracy_score(y_true, y_pred_classes) * 100, 3)} %")
print(f"Precision: {round(precision_score(y_true, y_pred_classes, average='weighted') * 100, 3)} %")
print(f" F1-Score: {round(f1_score(y_true, y_pred_classes, average='weighted') * 100, 3)} %")
print(f" Recall: {round(recall_score(y_true, y_pred_classes, average='weighted') * 100, 3)} %")
precision recall f1-score support
0 0.96 0.96 0.96 528
1 0.94 0.95 0.95 561
2 0.90 0.91 0.91 538
3 0.86 0.89 0.87 582
4 0.93 0.88 0.91 550
accuracy 0.92 2759
macro avg 0.92 0.92 0.92 2759
weighted avg 0.92 0.92 0.92 2759
Accuracy: 91.7 %
Precision: 91.758 %
F1-Score: 91.712 %
Recall: 91.7 %
In [22]:
plt.figure(figsize=(12, 5))
# График точности
plt.subplot(1, 2, 1)
plt.plot(trained_model.history['accuracy'], label='train_accuracy')
plt.plot(trained_model.history['val_accuracy'], label='val_accuracy')
plt.title('Accuracy over Epochs')
plt.xlabel('Epochs')
plt.ylabel('Accuracy')
plt.legend()
# График потерь
plt.subplot(1, 2, 2)
plt.plot(trained_model.history['loss'], label='train_loss')
plt.plot(trained_model.history['val_loss'], label='val_loss')
plt.title('Loss over Epochs')
plt.xlabel('Epochs')
plt.ylabel('Loss')
plt.legend()
plt.tight_layout()
# plt.savefig('neural_network_training_history.png', dpi=300, bbox_inches='tight')
plt.show()
# plot_model(model, to_file='model_architecture.png', show_shapes=True, show_layer_names=True)
In [23]:
# Построение матрицы ошибок для нейронной сети
confMat = ConfusionMatrixDisplay(confusion_matrix=confusion_matrix(y_true, y_pred_classes), display_labels=np.arange(5))
confMat.plot(cmap=plt.cm.Blues)
plt.title('Матрица ошибок для Нейронной сети')
# plt.savefig('neural_network_confusion_matrix.png', dpi=300, bbox_inches='tight')
plt.show()
